John W. Krakauer

Getting neurorehabilitation right: what can be learned from animal models?

Animal models suggest that a month of heightened plasticity occurs in the brain after stroke, accompanied by most of the recovery from impairment. This period of peri-infarct and remote plasticity is associated with changes in excitatory/inhibitory balance and the spatial extent and activation of cortical maps and structural remodeling. The best time for experience and training to improve outcome is unclear. In animal models, very early (<5 days from onset) and intense training may lead to increased histological damage. Conversely, late rehabilitation (>30 days) is much less effective both in terms of outcome and morphological changes associated with plasticity. In clinical practice, rehabilitation after disabling stroke involves a relatively brief period of inpatient therapy that does not come close to matching intensity levels investigated in animal models and includes the training of compensatory strategies that have minimal impact on impairment. Current rehabilitation treatments have a disappointingly modest effect on impairment early or late after stroke. Translation from animal models will require the following: (1) substantial increases in the intensity and dosage of treatments offered in the first month after stroke with an emphasis on impairment; (2) combinational approaches such as noninvasive brain stimulation with robotics, based on current understanding of motor learning and brain plasticity; and (3) research that emphasizes mechanistic phase II studies over premature phase III clinical trials.

The uses and interpretations of the motor-evoked potential for understanding behaviour

AbstractnThe motor-evoked potential (MEP) elicited in peripheral muscles by transcranial magnetic stimulation (TMS) over human motor cortex is one of the hallmark measures for non-invasive quantification of cortical and spinal excitability in cognitive and clinical neuroscience. In the present article, we distinguish three main uses for MEPs in studies of behaviour: for understanding execution and performance of actions, as markers of physiological change in the motor system, and as read-out of upstream processes influencing the motor system. Common to all three approaches is the assumption that different experimental manipulations act on the balance of excitatory and inhibitory pre-synaptic (inter)neurons at the stimulation site; this in turn contributes to levels of (post-synaptic) excitability of cortico-spinal output projections, which ultimately determines the size of MEPs recorded from peripheral muscles. We discuss the types of inference one can draw from human MEP measures given that the detailed physiological underpinnings of MEPs elicited by TMS are complex and remain incompletely understood. Awareness of the different mechanistic assumptions underlying different uses of MEPs can help inform both study design and interpretation of results obtained from human MEP studies of behaviour.

The proportional recovery rule for stroke revisited

Most motor recovery occurs in the first 3 months after stroke. This is often referred to as spontaneous biological recovery (SBR), a process that has been both underinvestigated in humans and underexploited with respect to neurorehabilitation. Data from animal models suggest that SBR is attributable to a unique timelimited period of enhanced postischemic plasticity. In 2008, we defined recovery as the difference (D) between impairment in the few days after stroke and at a later time point (3 months). We reasoned that SBR, being a process, is best reflected in a change rather than in a final endpoint. Our original finding was that SBR, in the majority of patients, follows a proportional recovery rule. Impairment was measured with the Fugl-Meyer Scale (FM), which assesses the ability of the patient to move individual joints out of synergy and therefore captures both motor control and strength. For the upper limb, the maximal score is 66. The proportional recovery rule states that, at 3 months, patients should get approximately 70% of their maximum potential recovery back. So, for example, a patient with moderate hemiparesis of 46 will recover (66-46) 3 0.7, which equals 60. This rule has since been validated for motor recovery in two subsequent studies, and we have shown that it appears to also be true for aphasia recovery. Interestingly, a subset of patients with severe hemiparesis (FM <20) fails to show proportional recovery. That is to say, whereas some patients with severe hemiparesis recover proportionally as all others, some are nonrecoverers. In contrast, patients with mild-to-moderate hemiparesis always recover to nearly the same degree. Thus these two findings—the rule itself and its failure in a subset of patients with severe hemiparesis—led us to two conclusions: (1) The existence of the rule implies that current rehabilitation methods in the first 3 months after stroke have little or no impact on recovery from impairment above what is expected from SBR, and (2) there is something categorically different between severe patients who do recover and those who do not. Specifically, we conjectured that recovery requires reorganization, which takes time, but this reorganization ultimately requires access to muscles through the corticospinal tract (CST). If the CST is interrupted too much, then no amount of cortical reorganization will make a difference. Accordingly, we predicted that the dichotomy between recoverers and nonrecoverers might map onto patients who do or do not have measurable motor-evoked potentials (MEPs) using transcranial magnetic stimulation (TMS). In two studies published in this issue of Annals of Neurology, two approaches have been taken to find CST-based predictors of recovery versus nonrecovery in patients with severe hemiparesis. Before describing the two studies, it is important to first examine the proportional recovery rule itself a little more carefully given that there are potential concerns when the same measure, initial impairment (FM0), is correlated with itself plus another term, final impairment (FM1), that is, the relationship between FM0 and FM1 2 FM0. Measuring FM0 and FM1 with error, FM0 1 e0 and FM1 1 e1, can induce positive correlations between initial impairment and the change in impairment, even when there is either no true recovery or true recovery that is unrelated to initial impairment. This correlation arises from the appearance of the error e0 in measured initial impairment, FM0 1 e0, and in the measured change in impairment, D 5 (FM1 1 e1) 2 (FM0 1 e0). However, the induced correlation will be small when the variability in true initial FM0 impairment is large compared to the variance of the measurement error e0, as is the case for FM, which has good reliability. Given low measurement error variance, it is reasonable to interpret D as true biological change. In contrast, high correlations between FM1 and FM0 are expected whether or not D is related to initial impairment. Although this correlation is not spurious, given that it accurately reflects the fact that patients with lower-than-average initial FM will have lower-than-average final FM, it does not directly address recovery. For this reason, we, and others subsequently, prefer to model D rather than FM1. Finally, when using measurements that have a constrained range, it is important to consider ceiling effects. Thus far, such effects have not been observed: Substantial room for improvement remains for all but the least affected subjects. A related concern is that FM measures latent “true recovery” nonlinearly. For example, a change from FM0 5 56 to FM1 5 61 (D 5 5) may reflect the same degree of “recovery” as a change from FM0 5 36 to FM1 5 51 (D 5 15). In this hypothetical scenario, true recovery is constant (and therefore unrelated to initial impairment), but the observed data are consistent with proportional recovery. This possibility, however,

Corticospinal excitability is enhanced after visuomotor adaptation and depends on learning rather than performance or error

We used adaptation to high and low gains in a virtual reality setup of the hand to test competing hypotheses about the excitability changes that accompany motor learning. Excitability was assayed through changes in amplitude of motor evoked potentials (MEPs) in relevant hand muscles elicited with single-pulse transcranial magnetic stimulation (TMS). One hypothesis is that MEPs will either increase or decrease, directly reflecting the effect of low or high gain on motor output. The alternative hypothesis is that MEP changes are not sign dependent but rather serve as a marker of visuomotor learning, independent of performance or visual-to-motor mismatch (i.e., error). Subjects were required to make flexion movements of a virtual forefinger to visual targets. A gain of 1 meant that the excursions of their real finger and virtual finger matched. A gain of 0.25 (low gain) indicated a 75% reduction in visual versus real finger displacement, a gain of 1.75 (high gain) the opposite. MEP increases (>40%) were noted in the tonically activated task-relevant agonist muscle for both high- and low-gain perturbations after adaptation reached asymptote with kinematics matched to veridical levels. Conversely, only small changes in excitability occurred in a control task of pseudorandom gains that required adjustments to large errors but in which learning could not accumulate. We conclude that changes in corticospinal excitability are related to learning rather than performance or error.

Explicit knowledge enhances motor vigor and performance: motivation versus practice in sequence tasks

Motor skill learning involves a practice-induced improvement in the speed and/or accuracy of a discrete movement. It is often thought that paradigms involving repetitive practice of discrete movements performed in a fixed sequence result in a further enhancement of skill beyond practice of the individual movements in a random order. Sequence-specific performance improvements could, however, arise without practice as a result of knowledge of the sequence order; knowledge could operate by either enabling advanced motor planning of the known sequence elements or by increasing overall motivation. Here, we examined how knowledge and practice contribute to performance of a sequence of movements. We found that explicit knowledge provided through instruction produced practice-independent improvements in reaction time and execution quality. These performance improvements occurred even for random elements within a partially known sequence, indicative of a general motivational effect rather than a sequence-specific effect of advanced planning. This motivational effect suggests that knowledge influences performance in a manner analogous to reward. Additionally, practice led to similar improvements in execution quality for both known and random sequences. The lack of interaction between knowledge and practice suggests that any skill acquisition occurring during discrete sequence tasks arises solely from practice of the individual movement elements, independent of their order. We conclude that performance improvements in discrete sequence tasks arise from the combination of knowledge-based motivation and sequence-independent practice; investigating this interplay between cognition and movement may facilitate a greater understanding of the acquisition of skilled behavior.

Persistent Residual Errors in Motor Adaptation Tasks: Reversion to Baseline and Exploratory Escape

When movements are perturbed in adaptation tasks, humans and other animals show incomplete compensation, tolerating small but sustained residual errors that persist despite repeated trials. State-space models explain this residual asymptotic error as interplay between learning from error and reversion to baseline, a form of forgetting. Previous work using zero-error-clamp trials has shown that reversion to baseline is not obligatory and can be overcome by manipulating feedback. We posited that novel error-clamp trials, in which feedback is constrained but has nonzero error and variance, might serve as a contextual cue for recruitment of other learning mechanisms that would then close the residual error. When error clamps were nonzero and had zero variance, human subjects changed their learning policy, using exploration in response to the residual error, despite their willingness to sustain such an error during the training block. In contrast, when the distribution of feedback in clamp trials was naturalistic, with persistent mean error but also with variance, a state-space model accounted for behavior in clamps, even in the absence of task success. Therefore, when the distribution of errors matched those during training, state-space models captured behavior during both adaptation and error-clamp trials because error-based learning dominated; when the distribution of feedback was altered, other forms of learning were triggered that did not follow the state-space model dynamics exhibited during training. The residual error during adaptation appears attributable to an error-dependent learning process that has the property of reversion toward baseline and that can suppress other forms of learning.

A motor planning stage represents the shape of upcoming movement trajectories.

Interactions with our environment require curved movements that depend not only on the final position of the hand but also on the path used to achieve it. Current studies in motor control, however, largely focus on point-to-point movements and do not consider how movements with specific desired trajectories might arise. In this study, we examined intentionally curved reaching movements that navigate paths around obstacles. We found that the preparation of these movements incurred a large reaction-time cost. This cost could not be attributed to nonmotor task requirements (e.g., stimulus perception) and was independent of the execution difficulty (i.e., extent of curvature) of the movement. Additionally, this trajectory representation cost was not observed for point-to-point reaches but could be optionally included if the task encouraged consideration of straight trajectories. Therefore, when the path of a movement is task relevant, the shape of the desired trajectory is overtly represented as a stage of motor planning. This trajectory representation ability may help explain the vast repertoire of human motor behaviors.

Motor Control of the Hand Before and After Stroke

The combination of the unique biomechanical properties of the human hand with the anatomy and physiology of the motor cortex and its descending pathways lead to an unprecedented behavioral repertoire. A prominent feature in the taxonomy of hand function is the distinction between power grip and finger individuation, which represent extremes along the continuous dimensions of stability and precision. Two features of the human central nervous system (CNS) are considered critical to hand dexterity: the distributed finger representation in primary motor cortex and its monosynaptic cortico-motoneuronal (CM) connections. The subset of motor cortical neurons, which contribute these CM connections, has been described as “new” M1, owing to its emergence in primates. Here we argue that the neural basis of hand impairment after stroke and its recovery can be attributed to the interplay between spared corticospinal projections and cortical modulation of brainstem descending pathways, specifically the reticulospinal tract.

The Future of Stroke Treatment Bringing Evaluation of Behavior Back to Stroke Neurology

Acute stoke interventions and stroke rehabilitation are aimed at salvaging or restoring brain function. How do we know if we have accomplished this goal? We examine the patient. One neurological historian asserted, “Most of the modern neurological examination evolved in a short time span, between 1850 and 1914 …”1. This quote is telling; it implies that the examination itself has not changed much since about 1900. For generations of medical students, residents, and other trainees in neurology, the neurological examination has achieved almost sacred, untouchable status, while at the same time becoming less important, as diagnostic technologies have become ever more sophisticated. Indeed, many of the examinations components have become almost empty ritual. Ask a resident what modern neuroscience has revealed about the mechanisms of, for example, increased tone, neglect, apraxia, and alexia, and how this new knowledge relates to the components of the neurological examination, or how the examination might be updated; you will likely be met with a blank stare. So ironically, even as cognitive neuroscience has advanced, the interest of neurologists in behavior in the broadest sense, and its underlying physiology and anatomy has waned. Thus, current stroke neurologists have largely failed to emphasize the evaluation of the effects of our interventions on brain function.

A Comparison of Two Methods for MRI Classification of At-Risk Tissue and Core Infarction

Objective: To compare how at-risk tissue and core infarction were defined in two major trials that tested the use of MRI in selecting acute stroke patients for endovascular recanalization therapy. Methods: MRIs from 12 patients evaluated for possible endovascular therapy were processed using the methods published from two major trials, MR RESCUE and DEFUSE 2. Specifically, volumes of at-risk tissue and core infarction were generated from each patient’s MRI. MRIs were then classified as whether or not they met criteria for salvageable tissue: “penumbral pattern” for MR RESCUE and/or “target profile” for DEFUSE 2 as defined by each trial. Results: Volumes of at-risk tissue measured by the two definitions were correlated (pu2009=u20090.017) while the volumes of core infarct were not (pu2009=u20090.059). The volume of at-risk tissue was consistently larger when defined by the penumbral pattern than the target profile while the volume of core infarct was consistently larger when defined by the target profile than the penumbral pattern. When these volumes were used to classify the MRI scans, 9 out of 12 patients (75%) were classified as having a penumbral pattern, while only 4 out of 12 patients (33%) were classified as having a target profile. Of the 9 patients classified as penumbral pattern, 5 (55%) were classified differently by the target profile. Interpretation: Our analysis found that the MR RESCUE trial defined salvageable tissue in a way that made it more likely for patients be labeled as favorable for treatment. For the cohort of patients examined in this study, had they been enrolled in both trials, most of the patients identified as having salvageable tissue by the MR RESCUE trial would not have been considered to have salvageable tissue in the DEFUSE 2 trial. Caution should be taken in concluding that MRI selection for endovascular therapy is not effective as imaging selection criteria were substantially different between the two trials.